Google DeepMind Workers Seek Union Recognition, Company Refuses

Employees at Google DeepMind's London office, one of the world's leading artificial intelligence labs, are trying to form a union. It is the first such attempt at a top-tier AI lab, and the stakes go beyond pay or holidays to who the company sells its technology to.

The vote and workers' demands

In May 2026, members of the Communication Workers Union (CWU) held a vote in which 98 percent backed forming a union. The initiative covers roughly a thousand employees at DeepMind's London office. Together with Unite the Union, workers sent a formal letter to Debbie Weinstein, Google's managing director for the UK and Ireland, demanding recognition of both unions as staff representatives.

Formally, the dispute concerns employment terms, pay, working hours and holidays. In practice, the real driver of the organizing effort is opposition to the military's use of DeepMind's technology. Workers also want Google to reinstate a pledge the company quietly removed from its website in February 2025, which had banned developing AI for weapons or surveillance that violates international norms.

Military contracts in the background

The immediate trigger for the unrest was Google's agreement with the US Department of Defense, allowing Gemini models to be used on classified military networks for any lawful purpose. Critics worry that such broad wording opens the door to developing autonomous weapons and mass surveillance with limited oversight.

On top of that sits Project Nimbus, a $1.2 billion contract supplying cloud services to the Israeli government and military. More than 600 Google employees signed an open letter opposing the Pentagon deal. Notably, OpenAI, xAI, Nvidia, Microsoft and Amazon have all signed similar military contracts, while only Anthropic has refused to cooperate, a decision that led the Pentagon to formally flag the company as a supply chain risk, a designation Anthropic is now challenging in court.

Google's response and mediation

Google has refused to voluntarily recognize the unions for collective bargaining. Instead, it proposed talks through Acas, the UK's state conciliation service, opening roughly a twenty day negotiating window. If mediation fails, the unions can refer the case to the Central Arbitration Committee, the body that rules on union recognition disputes in the UK.

John Chadfield, a national CWU representative, called the mediation offer a step forward, while stressing that workers' underlying concerns remain unchanged. In an official statement, Google acknowledged only that it had received the unions' letter and said it has always valued constructive dialogue with employees.

What it means for the AI industry

The situation echoes the 2018 dispute over Project Maven, when thousands of Google employees forced the company to abandon a military contract for analyzing drone footage. That employee pressure campaign succeeded, but in 2019 Google also disbanded its AI ethics council amid waves of protest, showing the limits of management's tolerance for internal dissent.

For the Polish market, the DeepMind dispute signals that tension between lucrative military contracts and the ethical commitments of AI labs is no longer a purely theoretical issue. If Acas mediation fails and the case reaches the Central Arbitration Committee, the outcome could set a precedent for worker organizing at other AI companies, including those operating in Europe.

Sources: Fortune (fortune.com), ResultSense (resultsense.com).